First, we want to visualize cancer cases per 100K people by county. We observe that data for Minnesota and Kansas are missing. Ignoring missing data, we see that Kentucky and New York have high cancer rates evenly distributed across all counties.

Then, we want to understand what attributes to high cancer rates. This second graph maps US Power Plants on top of the cancer rate map. The power plants are color coded by the primary type of fuel they utilize. We observe that power plants geographically overlap with counties with higher cancer rates. Some power plants, such as solar fuel plans, are only centered around a small area in the US. Click on the left button to dive into only one type of power plants. Hover over the power plants or counties to see their detailed information.

It is visually confusing to have so many colors on the same plot. Therefore, we simplify the power plants fuel types to either renewable or non-renewable. This third plot is color-coded by power plant renewability, and the plant size on the map is proportional to the output production in 2017 (Gigawatt hours). We see that renewable energy is at much smaller scale than non-renewable energy. Some of the highest cancer rate regions such as Kentucky is surrounded by a ring of high-output non-renewable energy plants.

What about toxic water sources? Could this be another factor that contributes to regional cancer spikes? This plot above shows the toxic water sources, color-coded by whether the source is federal or non-federal. Although Kentucky and New York are not occupied by large bubbles of toxic water sources, we see that regions with low cancer rates in the mid-west (excluding the west coast) coincidentally have almost no toxic water sources.

Do we have quantitative evidence to support these geographical observations? The last plot shows cancer rate per county with respect to four other factors: 1) the number of toxic water sources in the county, 2) the number of power plants in the county, 3) the toxic water score, and 4) the power plant production output in Gigawatt hours. Looking at the top two plots, we see that the number of toxic water sources and the number of power plants per county is not normally distributed. Although counties with less toxic water sources and less power plants have varying cancer rates from 276 to more than 550, we see that higher number of toxic water sources and higher number of toxic plants necessarily leads to higher cancer rate. Looking at the last two plots, we see that the water scores and power plant output (GWH) are more uniformly distributed, we observe the same patter in plot 3, where higher toxic water score necessarily sets the cancer rate lower threshold much higher than the minimum.